UIUC Physics 406 Acoustical Physics of Music

Mathematical Musical Physics of the Wave Equation

The purpose of this particular set of lecture notes for this course is to investigate the
mathematical physics (and the use) of the wave equation for describing wave behavior
associated with different kinds of one, two and three-dimensional physical systems – which
have relevance for various kinds of musical instruments. The wave equation mathematically
describes the behavior of waves for a given physical system, and is “generically” given by:

 1  2 (r , t )
  (r , t )  2
2
0

v t 2
where  (r , t ) is the displacement amplitude of the wave at the (1-, 2-, or 3-D) space position,

r at time, t from its equilibrium position; the symbol v represents the longitudinal speed of
propagation of the wave and  2 is the Laplacian operator (the form of which is relevant for the
dimensionality and symmetry of the physical system under investigation). Formally-speaking,
the above wave equation is a linear, homogeneous 2nd-order differential equation.
For musical instrument applications, we are specifically interested in standing wave
solutions of the wave equation (and not so much interested in investigating the traveling wave
solutions). We have discussed the mathematical physics associated with traveling and standing
waves in previous lecture notes for this course. Mathematically, standing wave solutions of the
wave equation – are formally known as eigen solutions, eigen modes, and/or also known as
normal modes – and such solutions result as a consequence of imposing specific boundary

conditions in space on either the amplitude of the wave, (r , t ) – which are formally known as
Dirichlet boundary conditions, or imposing specific boundary conditions on the spatial 1st
 
derivatives of  (r , t ) , e.g.  (r , t ) / x |x  xo – which are formally known as Neumann boundary

conditions, or imposing a combination of both – i.e. mixed boundary conditions on  (r , t ) and

spatial 1st derivatives of  (r , t ) – which are formally known as Cauchy boundary conditions.

Since we are specifically interested in standing wave eigen-solutions of the wave equation,
we will “go for the jugular” in these lecture notes and dispense with discussion of the most
general possible solutions of the wave equation for a given physical problem, in order to keep
the length of these lecture notes tractable. The reader is requested to bear this in mind!
The reader should also note that the above linear, homogeneous wave equation is a
mathematical description/treatment of wave phenomena at a first-order level – actual (i.e. real)
physical systems have additional physical processes that are simultaneously operative, e.g.
dissipative processes – such as internal/external friction and/or damping - air viscosity, various
other energy loss mechanisms, etc. and e.g. finite stiffness effects of vibrating systems, as
opposed to the perfectly compliant material implicitly assumed in the (derivation of the) above
wave equation. All these “higher-order” effects are (for the time being) temporarily neglected
here – usually these effects are small, resulting in perturbations/small (but easily
measurable/detectable) shifts in the normal modes of vibration of the mechanical system.
When such higher-order effects are explicitly included, they add/contribute additional
mathematical terms to the wave equation. Thus solution(s) to the more-sophisticated/realistic
wave equation are correspondingly more complicated and tedious to obtain.

1
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Furthermore, implicit in the mathematics of the above wave equation is the tacit assumption
that the physical material(s) comprising the various systems we are about to discuss are linear,
homogeneous and isotropic materials. There is nothing in the above form of the wave equation
that explicitly takes into account the proper description of non-linear, non-homogeneous and/or
anisotropic materials. The above wave equation is also really only valid strictly in the small-
amplitude-of-oscillations/vibrations regime. However, modifications to explicitly include such
effects can be incorporated into the wave equation to accommodate these additional properties
of such physical systems, if needed. Again, the solution(s) to the more-sophisticated/realistic
wave equation are correspondingly more complicated and tedious to obtain.
Here in these lecture notes, we wish to see the overall physics “forest”, and thus temporarily
neglect/ignore (some of) the details – the physics “trees”.

A. Standing Waves In One-Dimensional Systems:

A1. Transverse Standing Waves on a Vibrating String – Fixed Ends:
One-dimensional wave behavior on a vibrating string is mathematically described by the
1-D wave equation:
 2 ( x, t ) 1  2 ( x, t )
 2 0

x 2 v t 2 
where (r , t ) is the instantaneous transverse displacement amplitude of the string at the point r
at the time t. We can (trivially) rewrite this as:

 2 ( x, t ) 1  2 ( x, t )
 2
x 2 v t 2
Notice that the LHS of this equation has only space-derivatives of the one-dimensional
variable, x associated with it, whereas the RHS of this equation has only time-derivatives
associated with it. This suggests we try a product solution for the wave equation, i.e.
 ( x, t )  U ( x)T (t ) where U ( x) contains only spatially x-dependent terms and T (t ) contains
only temporally- (i.e. time)-dependent terms.
Formally, we insert  ( x, t )  U ( x)T (t ) into the above differential equation, and then
explicitly carry out the differentiation:
 2U ( x )T (t ) 1  2U ( x)T (t )
 2
x 2 v t 2
 2U ( x) 1  2T (t )
T (t )  2 U ( x)
x 2 v t 2
d 2U ( x) 1 d 2T (t )
T (t )  2 U ( x)
dx 2 v dt 2
1 d 2U ( x) 1 1 d 2T (t )
 2  k 2
U ( x) dx 2
v T (t ) dt 2

2
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

In the second line above, we’ve explicitly used the fact(s) that T (t ) and U ( x) , respectively
are functions of time, t and space, x only. On the third line, again since U ( x) and T (t ) ,
respectively are functions only of time and space, we formally convert partial derivatives of x
and t to total derivatives of x and t, respectively. Finally, on the fourth line, we divide both
sides of this equation by  ( x, t )  U ( x)T (t ) and then notice that the LHS is now a entirely
function of U ( x) only and that the RHS is now entirely a function of T (t ) . This last equation
must be satisfied for any/all possible values of x and t, and the only way this can happen is if
both sides of this latter equation are equal to a constant, which we (deliberately chose to) set
equal to –k2 (since we know what’s going to happen next…:). We call the constant that arises
from use of the separation of variables technique, the separation constant. Thus, here –k2 is the
separation constant.
Using the separation of variables technique, we actually wind up with two differential/wave
equations:
d 2U ( x)
  k 2U ( x)
dx 2

1 d 2T (t )
  k 2T (t )
v dt
2 2

Now v = f  = (/2)(2/k) =  / k; thus vk = . Rewriting the above two equations:

d 2U ( x)
 k 2U ( x)  0
dx 2

d T (t )
2
  2T (t )  0
dt 2

Both of these equations are of the same mathematical form – both are indeed wave equations.
The space-domain version of this linear, homogeneous 2nd –order differential equation is
known as the Helmholtz equation.
Thus far, we have not explicitly discussed any particular solution(s) of these wave equations –
we know they are oscillatory in space and time, and again, we seek standing wave solutions.
On most stringed instruments, such as the violin, viola, cello, guitar, mandolin, piano, etc.
the ends of the string(s) – e.g. at the bridge and nut (headstock) end of a guitar are ideally
rigidly attached to the body of the instrument in some manner. The string(s) on the instrument
then have (active) length, L – the distance between the two fixed ends (this is known as the
scale length). Defining our coordinate system such that the x-axis coincides with the
equilibrium shape of the string, one end of the string at x = 0 (e.g. the bridge), the other end at x
= L (the nut/headstock on a guitar). Mathematically, the so-called boundary condition at the
ends of the string(s) is that the displacement amplitude is zero at the ends of the string,
i.e. ( x  0, t )   ( x  L, t )  0 for fixed ends, independent of time, t.

3
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Thus, this boundary condition is only relevant to the U(x)-wave equation, i.e. in reality the
boundary condition for fixed ends is U ( x  0)  U ( x  L)  0 . In general, there are only two
allowed spatially periodic standing wave solutions to this wave equation – either sin kx, or
cos kx. The above boundary condition for fixed ends allows only the sin kx-type solutions,
because sin 0 = 0, and sin kL = 0 if and only if kL = n, n = 1, 2, 3, 4, … Denoting kn = n/L,
we see that the allowed standing wave solution(s) to the U(x)-wave equation are of the form:

U n ( x)  An sin(kn x)  An sin(n x L)
We also see that the boundary condition(s) on the spatial U(x) standing wave solutions to the
U(x)-wave equation determine the frequencies of the eigen-modes/normal modes of vibration.
Since  = 2/k, then n = 2/kn = 2L/n = 2L/n, n = 1, 2, 3, 4, …Since f = v/, then fn = v/n
and thus fn = v/n = nv/2L, n = 1, 2, 3, 4, …
The lowest mode of vibration (known as the fundamental), is when n = 1; thus this vibrational
standing wave eigen-mode is also known as the first harmonic, with f1 = v/2L and 1 = 2L (i.e.
L = ½ 1) and corresponding standing wave solution U1 ( x)  A1 sin k1 x  A1 sin  x / L , as shown
in the figure below:

x
x=0 x=L

Excluding the endpoints, note that the first harmonic/fundamental has no nodes in its spatial
wave function, U1(x).

The next highest eigen-mode (n = 2) is known as the second harmonic (also known as the first
overtone) with f2 = 2v/2L = v/L and 2 = 2L/2 = L and corresponding standing wave solution
U 2 ( x)  A2 sin k2 x  A2 sin 2 x / L as shown below. Note that the second harmonic has one
node in its spatial wavefunction, U2(x) at x = L/2.
The next highest eigen-mode (n = 3) is known as the third harmonic (also known as the second

x
x=0 x=½L x=L

vertone) with f3 = 3v/2L and 3 = 3L/2, and corresponding standing wave solution
U 3 ( x)  A3 sin k3 x  A3 sin 3 x / L , which has two nodes in its spatial standing wave function,
U3(x) at x = L/3 and x = 2L/3; and so on… in general, there are (n  1) nodes for the spatial
standing wave function, Un(x).

4
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

We have not yet discussed the allowed solutions to the temporal Helmholtz equation, so let
us now turn our attention to this. As stated earlier, the physical boundary conditions on the
ends of the strings place no direct constraints on the allowed solutions, T(t) of the temporal
Helmholtz equation. Thus, both sin t and cos t solutions in principle are (or must be)
allowed.

Also, since spatial wavelengths and temporal frequencies are intimately related to each
other via f = v/, and hence also the eigen-frequencies, fn = v/n = nv/2L,  n = vkn = nv/L,
then the allowed solutions to the temporal Helmholtz equation, T(t)  Tn(t), with both sin nt
and cos nt type solutions allowed. Note that the eigen-frequencies for standing waves of a
string with fixed endpoints are integer-multiples of the lowest mode of vibration, fn = nf1,
n = 1, 2, 3, 4, 5, …. with f1 = v/2L and n = 1/n with 1 = 2L.

In actuality, the detailed shape of the vibrating string at some particular instant in time
(usually conveniently taken to be t = 0) – known as the “initial conditions” (not to be confused
with {spatial} boundary conditions!) actually determines/defines the relative amount(s) of the
allowed sin nt vs. cos nt type solutions – because this simply specifies, for each eigen-mode
what its phase is/how far it is along in its oscillation cycle at time t = 0. Temporal phase
information is “encrypted” into the allowed eigen-solutions, Tn(t) of the temporal Helmholtz
equation as follows:

Tn (t )  bn sin nt  cn cos nt

With the following constraints/conditions on the coefficients bn and cn:
1  bn  1
1  cn  1
bn2  cn2  1
Of course, one could equivalently write the allowed eigen-solutions of the temporal Helmholtz
equation simply as one or the other of the following forms:
Tn (t )  sin n t   n  ; n  tan 1  cn bn   cot 1  bn cn 
Tn (t )  cos n t   n  ; n  tan 1  bn cn   cot 1  cn bn 
 n   n  2
These relations can be obtained directly from the above Tn(t) = bn sin nt + an cos nt relation
using the trigonometric identities for sin(A+B) and cos(A+B), respectively.
We can also write the allowed eigen-solutions of the temporal Helmholtz equation in yet
another, equivalent form, using complex notation:

Tn (t )  e 
i nt n 

5
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

The complete eigen-function solution(s)  n ( x, t )  U n ( x)Tn (t ) for standing waves on a

vibrating string with fixed ends are of the form:
 n ( x, t )  U n ( x)Tn (t )
 An sin  kn x  bn sin nt   cn cos nt    An sin  n x / L  bn sin  n vt / L   cn cos  n vt / L  
 An sin  kn x  sin n t   n   An sin  n x / L  sin  n vt / L    n 
 An sin  kn x  cos n t   n   An sin  n x / L  cos  n vt / L    n 
i  n vt / L  n 
 An sin  kn x  e   An sin  n x / L  e 
i n t n 

with n = 2L/n and fn = v/n = nv/2L, n = 1, 2, 3, 4, 5, … and eigen-energies En  14 M n2 An2 .

The displacement amplitude coefficients, An are also formally specified/determined by the

initial conditions (i.e. harmonic content) of the vibrating string at time t = 0. Physically, this
means that the detailed shape/configuration of the vibrating string at time t = 0, e.g. a triangle,
sawtooth or square wave-shape, etc. completely specifies – by the method of Fourier analysis
(i.e. harmonic analysis) – the exact harmonic content (i.e. allowed fn values), the harmonic
amplitudes (values of An) and the phases, n (or n). For example, for a symmetrical triangle-
type standing wave (which has reflection symmetry about its mid-point), only odd-n
coefficients An are non-zero. For an asymmetrical triangle-type standing wave, which does not
have reflection symmetry about its mid-point) both even and odd-n coefficients An are non-
zero. For a 50% duty-cycle type square wave (which also has reflection symmetry about its
mid-point), again only odd-n coefficients An are non-zero. For further details of how this is
accomplished, see e.g. the UIUC P498POM lecture notes on Fourier Analysis, I-IV.
As mentioned at the outset of this section, the above eigen-function solutions
 n ( x, t )  U n ( x)Tn (t ) for standing waves on a vibrating string with (idealized) fixed ends are
relevant for a broad selection of stringed instruments, such as the violin, viola, cello, guitar,
mandolin, piano, etc.
A2. Transverse Standing Waves on a Vibrating String – Free Ends:
Certain kinds of 1-dimensional systems with free ends (Neumann boundary conditions) can
also exhibit standing wave solutions. Mathematically, the method of analysis is exactly the
same as above, except that for free, rather than fixed ends – i.e. Neumann boundary conditions,
the value of the spatial derivative (= slope) of U(x) must vanish at the endpoints x = 0 and x = L
for any/all time(s) t:
dU ( x) dU ( x)
 0 and 0
dx x 0 dx x  L

The reader can easily verify that the allowed eigen-solutions Un(x) for free ends e.g. of a
vibrating string, or a transversely-vibrating 1-dimensional rod must be of the form cos knx
(rather than sin knx, as in the case of fixed ends/Dirichlet boundary conditions). The form of the
temporal eigen-solutions, Tn(t) are the same for free ends/Neumann boundary conditions and
for fixed ends/Dirichlet boundary conditions.
6
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Physically, we can envision an ideal stretched string of length, L with free ends, tension, T
and mass per unit length  = M/L as having massless, frictionless rings attached to the ends of
the string, each sliding frictionlessly on a perpendicular rod that constrain the motion of the
ends of the string to be only in the transverse direction, as shown in the figures below.
The complete eigen-function solution(s)  n ( x, t )  U n ( x)Tn (t ) for standing waves on a
vibrating string with free ends are of the form:
 n ( x, t )  U n ( x)Tn (t )
 An cos  kn x  bn sin nt   cn cos nt    An cos  n x / L  bn sin  n vt / L   cn cos  n vt / L  
 An cos  kn x  sin nt   n   An cos  n x / L  sin  n vt / L    n 
 An cos  kn x  cos n t   n   An cos  n x / L  cos  n vt / L   n 
i  n vt / L  n 
 An cos  kn x  e   An cos  n x / L  e 
i n t n 

with n = 2L/n and fn = v/n = nv/2L, n = 1, 2, 3, 4, 5, … and eigen-energies En  14 M n2 An2 .

The first few standing wave eigen-modes for free ends on a string are shown below:

n=1
L=½
x
x=0 x=L

n=2
L=
x
x=0 x=L

n=3
L = 3/2 
x
x=0 x=L

Note the phase relation between relative motion of the displacement amplitude at the left
and right ends for odd vs. even n free-end eigen-modes. For odd n, the motion of the ends is
180o out of phase, for even n, the motion of the ends is in-phase.

7
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

A3. Transverse Standing Waves on a Vibrating String – Fixed End + Free End:
For completeness, we consider the possibility of standing waves on a vibrating 1-
dimensional system, such as a stretched string with mixed boundary conditions – i.e. so-called
Cauchy boundary conditions, where one end of the string (e.g x = 0) is fixed (thus U(x=0) = 0)
and the other end of the string (x = L) is free (U(x=L)/x = 0) (or vice-versa).
For the mixed boundary condition case, the reader can verify that fixed-free end boundary
conditions with a node at the fixed end (x = 0) and an anti-node at the free end (x = L) requires
spatial eigen-solutions of the form U n ( x)  An sin  kn x   An sin  2 x /   with n = 4L/n and
fn = v/n = nv/4L, with only odd-n integers allowed, i.e. n = 1, 3, 5, 7… etc. The complete
eigenfunction solutions for this case have the same form as that given in Section A1 above. For
free-fixed end conditions (U(x=0)/x = 0 and U(x= L) = 0), the spatial eigen-function
solutions are of the form U n ( x)  An cos  kn x   An cos  2 x / n  and thus the complete eigen-
function solutions have the same form as that given in Section A2 above, but again only odd-n
integers are allowed, i.e. n = 1, 3, 5, 7… and with eigen-energies En  14 M n2 An2 .

For either version of mixed fixed-free or free-fixed end conditions, the lowest mode – the
fundamental/first harmonic (n = 1) has 1 = 4L and f1 = v/4L; the second harmonic (n = 3) has
3 = 4L/3 and f3 = 3v/4L; the third harmonic (n = 5) has 5 = 4L/5 and f5 = 5v/4L, and so on.
The spatial waveforms for the first few/lowest eigen-modes for mixed, fixed-free end
conditions are shown in the figures below. The spatial waveforms for eigen-modes associated
with free-fixed end conditions are mirror-reflections of the fixed-free eigen-mode solutions.

L=¼
x
x=0 n=1 x=L

L = 3/4 
x
n=3 x=L
x=0

L = 5/4 
x
x=0 n=5 x=L

8
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

A4. Longitudinal Standing Waves on the Singing Rod – Free Ends:

The singing rod of length, L and diameter, D << L is a one-dimensional vibrating system
with eigen-modes that are longitudinal standing waves. The two ends of the rod at x = 0 and
x = L vibrate longitudinally (i.e. along the axis of the rod), thus we have free-free boundary
conditions at the rod ends, i.e. Neumann boundary conditions, zero slopes at the ends of the
rods, (U(x=0)/x = 0 and (U(x=L)/x = 0. The spatial eigen-solutions for the singing rod are
the same as given in Section A2 above, i.e. they are of the form
U n ( x)  An cos  kn x   An cos  2 x /   with n = 2L/n and fn = v/n = nv/2L, n = 1, 2, 3, 4, …
with eigen-energies En  14 M n2 An2 and longitudinal displacement waveforms for the first few
eigen-modes, also as shown in Section A2 above.
For a singing rod made up of an elastic solid, such as a metal, the longitudinal speed of
propagation of sound in the solid, vL is given by v  Y /  where Y (N/m2) = Young’s modulus
and  (kg/m3) is the density of the for the material. Physically, Young’s modulus, Y = / =
ratio of longitudinal compressive stress ( = F/A) to longitudinal strain ( = |L2-L1|/L1
{dimensionless}, where L1 is the equilibrium length of the rod, and L2 is the extended length of
the rod).
To make a singing rod “sing” in one of its eigen-modes, one grasps the rod at a
displacement node for that mode, e.g. with thumb and index finger, and then using the thumb
and index finger of the other hand, pull sharply along the axis of the rod, toward the end of the
rod, trying to stretch it. This works best coating the pulling thumb & index fingers first with
crushed violin rosin, in order to really get a good “pull” on the rod. A typical displacement
amplitude for the fundamental, f1 ~ 1670 Hz on a L ~ 1.5 m, ~ 1” diameter aluminum rod is
(x) ~ 1 mm. Its also very loud, with measured sound pressure level of ~ 140 dB at the rod
ends! For more information specifically on the singing rod, see the UIUC P498POM lecture
notes on the singing rod.

A5. Longitudinal Standing Waves on a Stretched String or a Thin Bar:

Longitudinal modes of vibrations of a stretched string, or a thin bar, while much less
common than transverse standing waves in these same systems, do occur in certain
circumstances. However, unlike transverse waves on a string or a thin bar, the longitudinal
speed of propagation (and hence their eigen-frequencies) do not change with tension (unless
the physical properties of the string or thin bar change with tension), since vL  Y /  . For
both fixed ends or both free ends, the eigen-solutions  n ( x, t )  U n ( x)Tn (t ) for longitudinal
displacement are as given in Section A1 or A2 above, respectively, both situations have n =
2L/n and fn = vL/n = nvL/2L, n = 1, 2, 3, 4, 5, … and eigen-energies En  14 M n2 An2 . For mixed
fixed-free boundary conditions, the eigen-solutions for longitudinal displacement
 n ( x, t )  U n ( x)Tn (t ) are as given in Section A3 above, with n = 4L/n and fn = vL/n = nvL/4L,
with only odd-n integers allowed, i.e. n = 1, 3, 5, 7… and with eigen-energies En  14 M n2 An2 .

9
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

A6. Longitudinal Standing Waves in Long, Narrow Organ Pipes:

The standing waves that exist in long, narrow organ pipes (diameter, D << length, L) of
various kinds – closed-closed, open-open and closed-open or open-closed end conditions are
excited by air flow – hence air pressure supplies the needed energy input, rather than
mechanical displacement.
Due to the relation between the air over-pressure amplitude, pn(x,t) and the longitudinal
displacement amplitude, n ( x, t ) , namely that pn ( x, t )   Bair d n ( x, t ) / dx where Bair is the
bulk modulus of air, the mathematical form of longitudinal displacement amplitude eigen-
solutions,  n ( x, t )  U n ( x)Tn (t ) for organ pipes with closed-closed end conditions (longitudinal
displacement nodes and overpressure antinodes at the closed ends of the organ pipe of length,
L) are as given above in Section A1, with n = 2L/n and fn = v/n = nv/2L, n = 1, 2, 3, 4, … and
eigen-energies En  14 M airn2 An2 , as shown in the figure below for the first few eigen-modes:

For each of the above (and following) organ pipe end-condition cases, the over-pressure
amplitude eigen-solutions Pn ( x, t ) can be obtained using the above relation between over-
pressure amplitude and longitudinal displacement amplitude.
For organ pipes with open-open end conditions (longitudinal displacement antinodes and
overpressure nodes at the open ends of the organ pipe of length, L) the mathematical form of
longitudinal displacement eigen-solutions,  n ( x, t )  U n ( x)Tn (t ) are as given above in Section
A2, again with n = 2L/n and fn = v/n = nv/2L, n = 1, 2, 3, 4, 5, … and eigen-energies
En  14 M airn2 An2 , as shown in the figure below for the first few eigen-modes:

10
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music
__________________________________________________________________________________________________________________

For organ pipes with open-closed or closed-open end conditions (longitudinal displacement
nodes (antinodes) and overpressure antinodes (nodes) at the closed (open) ends of the organ
pipe of length, L, respectively) the mathematical form of longitudinal displacement eigen-
solutions,  n ( x, t )  U n ( x)Tn (t ) are as given above in Section A3, with n = 4L/n and fn = v/n =
nv/4L, with only odd-n integers allowed, n = 1, 3, 5, 7… and eigen-energies En  14 M airn2 An2
as shown below for the first few eigen-modes of a closed-open organ pipe of length, L:

Note that an end correction exists – e.g. a last displacement node does not occur precisely at
the open end of the organ pipe, but is located a distance ~ D outside it. Hence a more accurate
wavelength formula is n'  n0 (1   D), ~ O(1)m 1 where the n0 are as given above.
11
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

B. Standing Waves In Two-Dimensional Systems:

We now turn our attention to standing waves associated with two-dimensional systems.
B1. Transverse Standing Waves on a Rectangular Membrane – Fixed Edges:
For a thin, perfectly compliant (i.e. flexible) rectangular membrane (e.g. an idealized
rectangular drum head) of dimensions Lx and Ly, the wave equation in rectangular 2-
dimensional coordinates (x,y) for the displacement amplitude,  ( x, y, t ) is given by:

1  2 ( x, y, t )
  ( x, y , t )  2
2
0
v t 2
where the longitudinal speed of propagation of transverse waves on a stretched, 2-dimensional
membrane is given by v  T /  ; T is the membrane surface tension (in Newtons/meter) and
  M/A = M /LxLy is the areal mass density of the membrane (in kg/m2). The Laplacian
operator, 2 in 2-D rectangular coordinates is given by:

2 2
2  
x 2 y 2

Thus, the 2-dimensional wave equation describing the behavior of waves on a rectangular
membrane is given by:

 2 ( x, y, t )  2 ( x, y, t ) 1  2 ( x, y, t )
  2 0
x 2 y 2 v t 2
Again, we can (trivially) rewrite this as:
 2 ( x, y, t )  2 ( x, y, t ) 1  2 ( x, y, t )
  2
x 2 y 2 v t 2
Notice again that the LHS (RHS) contains only spatial-dependent (time-dependent) functions,
respectively. Thus, we again can use the technique of separation of variables, with
 ( x, y, t )  U ( x, y )T (t ) where U ( x, y ) contains only spatially x- and y-dependent terms and
T (t ) contains only the time-dependent term.

Again, we have the relation v = f  = (/2)(2/k) =  / k; thus vk = . We again obtain a

separation constant of – k2, and, after some simple algebraic manipulations, obtain the
following two linear, homogeneous differential equations:
 2U ( x, y )  2U ( x, y )
  k 2U ( x, y )  0
x 2
y 2

d 2T (t )
  2T (t )  0
dt 2

12
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

We can again use the separation of variables technique on the above Helmholtz equation, with
a product solution of the form U ( x, y )  X ( x)Y ( y ) . Inserting this into the above Helmholtz
equation and carrying out the (partial) differentiations, dividing by U ( x, y )  X ( x)Y ( y ) and
carrying out a simple algebraic manipulation, we obtain the following equation:

1 d 2 X ( x) 1 d 2Y ( y )
  k 2
X ( x) dx 2 Y ( y ) dy 2

Since the first (second) term on the LHS depends only on x (y), respectively, in order for this
relation to be satisfied for all possible values of (x,y), each of these two terms must be equal to
a constant, which we call – kx2 and – ky2, respectively. Thus we obtain the following relation,
known as the so-called characteristic equation:

k 2  k x2  k y2
with:
d 2 X ( x)
 k x2 X ( x)  0
dx 2
d 2Y ( y )
 k y2Y ( y )  0
dy 2

With fixed-end/Dirichlet boundary conditions on both the X(x)- and Y(y)-solutions of:

X ( x  0)  X ( x  Lx )  0
Y ( y  0)  Y ( y  Ly )  0
we again obtain spatial eigen-mode solutions for 2-D transverse standing waves of the form:

X m ( x) ~ sin(km x)  sin(m x Lx )
Yn ( y ) ~ sin(kn y )  sin(n y Ly )
U m,n ( x, y )  X m ( x)Yn ( y )  Am,n sin(km x)sin(kn y )  Am,n sin(m x Lx )sin(n y Ly )
where m and n are both integers, i.e. m, n = 1, 2, 3, 4, 5, … Note that solutions with m = 0 or n
= 0 are not allowed, because then U m, n ( x, y )  X m ( x)Yn ( y )  0 everywhere, which are not
(propagating) transverse standing wave solutions.
The characteristic equation for transverse standing wave eigen-mode solutions for a 2-D
rectangular membrane and eigen-wavelengths become:

km2 ,n  km2  kn2   m / Lx    n / Ly  ;m,n  2 / km,n  2  m / Lx    n / Ly 

2 2 2 2

km   m / Lx ;m  1,2,3,4,5,...
kn   n / Ly ;n  1,2,3,4,5,...
13
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Since  = vk, with v  T /  the angular eigen-frequencies are m, n  vkm, n  v km2  kn2 and
thus eigen-frequencies fm,n and eigen-wavelengths  m,n are:

 m / Lx    n / Ly 
2
f m ,n  m ,n / 2  vkm ,n / 2  v / m , n  12 v
2

 m / Lx    n / Ly 
2
m ,n  2 / km,n  2
2

m, n  1,

The eigen-mode solutions of the associated temporal wave equation for two-dimensional
standing waves on a rectangular membrane are of the following equivalent form(s):

Tm ,n (t )  bm, n sin m,nt  cm ,n cos m ,nt

1  bm ,n  1 1  cm,n  1 bm2 ,n  cm2 ,n  1
Tm, n (t )  sin m ,nt   m, n  ; m, n  tan 1  cm,n bm ,n   cot 1  bm ,n cm,n 
Tm ,n (t )  cos m ,nt  m ,n  ;m ,n  tan 1  bm ,n cm ,n   cot 1  cm,n bm ,n 
 m,n   m,n  2
i m ,n t m ,n 
Tm ,n (t )  e
The complete eigen-mode solutions for two-dimensional transverse standing waves on a
rectangular membrane of dimensions with fixed edges are thus given e.g. by:
 m ,n ( x, y, t )  U m , n ( x, y )Tm , n (t )  X m ( x)Yn ( y )Tm ,n (t )
 m , n ( x, y , t )  Am ,n sin( km x) sin( kn y )e 
i  m , n t  m , n 

 m ,n ( x, y , t )  Am , n sin(m x Lx ) sin( n y Ly )e   i  m ,n t  m , n

With eigen-frequencies, eigen-wavelengths and eigen-energies and eigen-energies of:

 m / Lx    n / Ly 
2
f m, n  m ,n / 2  vkm, n / 2  v / m ,n  12 v
2

 m / Lx    n / Ly 
2
m ,n  2 / km ,n  2
2

Em, n  14 M m2 ,n Am2 ,n   2 Mf m2, n Am2 ,n  14  2 v 2 MAm2 ,n  m / Lx    n / Ly  

2 2

 
m, n  1,

14
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Note that the allowed standing wave eigen-solutions all have transverse displacement nodes
along the x- and y-edges of the rectangular membrane. The lowest frequency standing-wave
mode occurs when m = n = 1, with 1,1  2 1/ L2x  1/ L2y and f1,1  2v / 1/ L2x  1/ L2y and
E1,1  4 2 v 2 MA1,1
2
/ 1/ L2x  1/ L2y  . The exact sequence of which m,n eigen-values have
progressively higher frequencies depends on the detailed geometry of the rectangular plate,
whether Lx  Ly or Lx  Ly , or Lx  Ly or Lx  Ly .

Some of the lower-order eigen-modes of vibration for transverse standing waves on a

rectangular membrane are shown (with dashed nodal lines) in the figure below, for Lx  Ly :

m=n=1 m = 2, n = 1 m = 1, n = 2

m=n=2 m = 3, n = 1 m = 3, n = 2

For the special case of a square membrane, when Lx  Ly  L the square membrane has
exactly the same number of eigen-modes as that associated with a rectangular membrane
Lx  Ly , however the square membrane now has so-called 2-fold degeneracies associated with
it – i.e. distinct eigen-states,  m,n ( x, y, t ) with m  n (e.g.  1,2 ( x, y, t ) and  2,1 ( x, y, t ) ), but
which have the same eigen-frequencies, eigen-wavelengths and eigen-energies:

f m ,n  f n ,m  v / m ,n  m 2  n 2 v /  2 L 
m ,n  n ,m  2 L / m 2  n 2
Em ,n  En ,m  1
4L  2  m 2  n 2  v 2 MAm2 ,n

The 2-fold degeneracies arise because of the 2 spatial degrees of freedom (x & y) .AND. the
rotational symmetry of the square membrane – i.e. a square is invariant under 90o rotations.

15
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

B2. Transverse Standing Waves on a Circular Membrane – Fixed Edge:

We now consider standing waves on 2-dimensional system with circular symmetry – that of
a thin, perfectly compliant (i.e. flexible) circular membrane (e.g. an idealized circular drum
head) of radius R, the wave equation in cylindrical 2-dimensional coordinates (x,y  r,) for
the displacement amplitude,  (r ,  , t ) is given by:
1  2 (r ,  , t )
 2 (r ,  , t )  0
v2 t 2
Where the longitudinal speed of propagation of transverse waves on a stretched, 2-dimensional
circular membrane is (also) given by v  T /  where T is the membrane surface tension
(in Newtons/m) and   M/A = M /R2 is the areal mass density of the membrane (in kg/m2).
With x = rcos, y = rsin, d2r = rdrd, the Laplacian operator, 2 in cylindrical 2-D
coordinates is given by:
1     1 2
 
2
r 
r r  r  r 2  2
Thus, the 2-dimensional wave equation describing the behavior of waves on a cylindrical
membrane is given by:
 2 (r ,  , t ) 1  (r ,  , t ) 1  (r ,  , t ) 1  2 (r ,  , t )
  2  2 0
r 2 r r r  2 v t 2
Again, we can (trivially) rewrite this as:

 2 (r ,  , t ) 1  (r ,  , t ) 1  (r ,  , t ) 1  2 (r ,  , t )
  2  2
r 2 r r r  2 v t 2
Notice again that the LHS (RHS) contains only spatial-dependent (time-dependent) functions,
respectively. Thus, we again can use the technique of separation of variables, with
 (r ,  , t )  U (r ,  )T (t ) where U (r ,  ) contains only spatially r- and -dependent terms and
T (t ) contains only the time-dependent term.

Again, we have the relation v = f  = (/2)(2/k) = /k; thus vk = . We again obtain a

separation constant of – k2, and, after some simple algebraic manipulations, obtain the
following two linear, homogeneous differential equations:
 2U (r ,  ) 1 U (r ,  ) 1  2U (r ,  )
  2  k 2U (r ,  )  0
r 2
r r r  2

d 2T (t )
  2T (t )  0
dt 2

We can again use the separation of variables technique on the above spatial equation, with a
product solution of the form U (r ,  )  R(r )( ) . Inserting this into the above spatial equation
and carrying out the (partial) differentiations, dividing by U (r ,  )  R(r )( ) and carrying out
a simple algebraic manipulation, we obtain the following equation:
16
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

1  2 d 2 R(r ) dR(r )  1 d 2( )

r  r  k r
2 2
R r  
R(r )  
( )
dr 2 dr  ( ) d 2
The LHS (RHS) of this equation depends only on r (), respectively. Again, this can only
be true for all possible values of (r,), if both LHS and RHS are equal to a (dimensionless)
constant. We know that the ()-solutions must be periodic/singled valued (i.e. (=0) =
(=2)) or more generally, (=o) = (= o+2m)), m = 0, 1, 2, 3, 4, 5,….
Hence we will choose this separation constant to be m2.
The mathematical form of the standing-wave m() eigen-solutions we seek for modal
vibrations on a circular membrane must satisfy:

d 2  m ( )
 m2  m ( )  0
d 2
Thus, the m() eigen-solutions are (one of) the following two equivalent form(s):
 m ( )   m cos m   m sin m
1   m  1 1   m  1  m2   m2  1
 m ( )  e  ; m  tan 1   m  m 
i m  m 

m  0, 1, 2, 3,....

The radial equation is the well-known Bessel’s equation:

d 2 R(r ) 1 dR(r )  2 m2 
   k  2  R(r )  0
dr 2 r dr  r 
The most general solution for Bessel’s equation, with m = integer (which is the case we have
here) is of the form:
Rm (r )  Am J m (kr )  BmYm (kr )
J m ( x) { Ym ( x) } are the ordinary Bessel functions of order, m of the 1st {2nd} kind, respectively.

The J m ( x) are finite at x = 0 and are usually expressed as a power series expansion in x:

 1
r m 2 r

x
J m ( x)    
r  0 r ! ( m  r  1)  2 

Note that for m = integer, J  m ( x)  (1) m J m ( x) . The Ym ( x) can be expressed in various ways:
J m ( x) cos(m )  J  m ( x)
Ym ( x) 
sin(m )
For m = integer, these can also be written as:
1  J m ( x) m J  m ( x ) 
Ym ( x )     1
  m m 

17
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

The Ym ( x) are singular (they become (negative) infinite) at x = 0. However, because we used
cylindrical coordinates for our circular membrane, the origin (r = 0) is included in this
problem. Physically, we do NOT allow infinite amplitude displacements R(r)   for any
value of r, since an implicit initial assumption was small amplitude oscillations! Thus all of the
Bm coefficients for the Ym ( x) must be Bm = 0 for physically allowed eigen-mode solutions of the
2-D circular membrane.
Plots of the first few J m ( x) vs. x are shown in the figure below:

Plots of the first few Ym ( x) vs. x are shown in the figure below:

18
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

The radial boundary condition for transverse standing waves on a circular membrane with
fixed edge (i.e. zero transverse displacement) at r = R is Rm(r=R) = 0, i.e. Jm(kR) = 0. Since r =
R > 0, this means we seek the zeroes of Jm(kR), i.e. Jm(x=kR) = 0. Because of the complexity of
the form of the Jm(x), the zeroes of Jm(x) (and the Ym(x)) are non-analytic in nature, rather, they
are tabulated in many mathematical books, or they can be determined either by graphical
and/or computational numeric techniques. We summarize the first few zeroes of the low-order
Jm(x) in the table below:
n=1 n=2 n=3  zero #
m=0: J0(x)=0: x  2.40, 5.52, 8.65, ….
m=1: J1(x)=0: x  3.83, 7.02,10.17, ….
m=2: J2(x)=0: x  5.14, 8.42,11.62, ….

Since x = kR, then k = x/R and noting that again, for this 2-dimensional standing wave eigen-
value problem we have two indices, m and n to denote the eigen-wavenumbers km, n  xm ,n / R ,
and the eigen-frequencies m, n  vkm ,n  f m ,n  v / m ,n with v  T  . The eigen-energies,
Em ,n  14 M m2 ,n Am2 ,n and eigen-functions  m ,n (r ,  , t )  Rm,n (r ) m ( )Tm,n (t ) . The eigen-mode
solutions of the associated temporal wave equation for two-dimensional standing waves on a
circular membrane are of the following equivalent form(s):

Tm,n (t )  bm ,n sin m ,nt  cm ,n cos m ,nt

1  bm ,n  1 1  cm, n  1 bm2 ,n  cm2 ,n  1
Tm ,n (t )  sin m, nt   m ,n   cos m ,nt  m ,n  m ,n  m ,n  2
i m ,n t m ,n 
Tm ,n (t )  e
The complete eigen-mode solutions for two-dimensional standing waves on a circular
membrane of radius, R with fixed edges are thus given e.g. by:
 m , n ( r ,  , t )  Rm ,n ( r ) m ( )Tm ,n (t )
 m ,n ( r ,  , t )  Am , n J m ( km ,n R )e 
i m  m ,n  i m ,n t  m ,n 
e
 m ,n ( r ,  , t )  Am , n J m ( km ,n R )  m cos m   m sin m  bm ,n cos m ,n t  cm ,n sin m ,n t 

with eigen-frequencies, eigen-wavelengths and eigen-energies and eigen-energies of:

f m ,n  m ,n / 2  vkm ,n / 2  v / m, n m, n  2 / km ,n  2 R / xm,n Em ,n  14 M m2 ,n Am2 ,n
m  0,1,
 n  1,
The lowest modes of transverse standing waves on a circular membrane are listed below:
m=0, n=1: k0,1  2.40/R 0,1  2.40v/R  0,1(r,,t) = A0,1 J0(k0,1R) T0,1(t)
m=1, n=1: k1,1  3.83/R 1,1  3.83v/R  1,1(r,,t) = A1,1 J1(k1,1R) [1cos + 1sin]T1,1(t)
m=2, n=1: k2,1  5.14/R 2,1  5. l4v/R  2,1(r,,t) = A2,1 J2(k2,1R) [2cos2+2sin2]T2,1(t)
m=0, n=2: k0,2  5.52/R 0,2  5.52v/R  0,2(r,,t) = A0,2 J0(k0,2R) T0,2(t)
19
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Some of the lower-order eigen-modes of vibration for transverse standing waves on a circular
membrane (with dashed nodal lines) are shown in the figure below:

m =0, n =1 m =1, n =1 m =2, n =1

m+n=1; J0(k0,1r) m+n=2; J1(k1,1r)ei m+n=3; J2(k2,1r)e2i
(2-fold degenerate) (2-fold degenerate)

m =0, n =2 m =1, n =2 m =2, n =2

m+n=2; J0(k0,2r) m+n=3; J1(k1,2r)ei m+n=4; J2(k2,2r)e2i
(2-fold degenerate) (2-fold degenerate)

m =0, n =3 m =1, n =3 m =2, n =3

m+n=3; J0(k0,3r) m+n=4; J1(k1,3r)ei m+n=5; J2(k2,3r) e2i
(2-fold degenerate) (2-fold degenerate)

The 2-fold degeneracies for m > 0 again arise because of the 2 spatial degrees of freedom
(x and y, or r and ) .AND. the rotational symmetry of the circular membrane – it is invariant
under arbitrary rotations.

20
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.
 UIUC Physics 406 Acoustical Physics of Music

Legal Disclaimer and Copyright Notice:

Legal Disclaimer:

The author specifically disclaims legal responsibility for any loss of profit, or any consequential,
incidental, and/or other damages resulting from the mis-use of information contained in this document.
The author has made every effort possible to ensure that the information contained in this document is
factually and technically accurate and correct.

Copyright Notice:

The contents of this document are protected under both United States of America and
International Copyright Laws. No portion of this document may be reproduced in any manner
for commercial use without prior written permission from the author of this document. The
author grants permission for the use of information contained in this document for private,
non-commercial purposes only.

21
Professor Steven Errede, Department of Physics, University of Illinois at Urbana-
Champaign, Illinois 2002-2017. All rights reserved.